Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Demographic Predictors of Melanoma Stage at Diagnosis

Demographic Predictors of Melanoma Stage at Diagnosis ObjectiveTo examine sociodemographic characteristics as possible predictors of late-stage melanoma diagnosis. We hypothesized that late-stage diagnosis would be associated with the following: older age, male sex, unmarried status, lower educational attainment and income level, rural residence, and cigarette smoking.MethodsWe used data from the state tumor registry to study all incident cases of melanoma occurring in Florida during 1994 whose stage at diagnosis was available (N=1884). We used multiple logistic regression to determine the effects of sociodemographic characteristics on the odds of late-stage (regional or distant metastases) diagnosis.ResultsThere were 243 patients (12.9%) diagnosed as having melanoma that had metastasized to either regional lymph nodes or distant sites. Patients who were unmarried (odds ratio, 1.5; P=.01), male (odds ratio, 2.2; P<.001), or smokers (odds ratio, 2.2; P<.001) or who resided in communities with lower median educational attainment (odds ratio, 1.5; P=.048) had greater odds of having a late-stage diagnosis.ConclusionsTo detect these cancers at an earlier stage and improve outcomes, there should be increased educational efforts directed toward physicians who treat these patients. A recognition that there may be additional risk factors for late-stage diagnosis, beyond the established risk factors, such as family history and excess sun exposure, should be included in the initial assessment. Specific public education efforts should also be targeted to these patients to increase their self-surveillance and surveillance of their partners.DURING 1973-1989, the incidence of malignant melanoma increased 85%, more than any other major cancer.There were approximately 42,000 new cases in 1998, with 7000 deaths.The lifetime risk of acquiring melanoma is now estimated to be about 1 in 87.Within the next 10 years, the risk is expected to climb to 1 in 75.The prognosis of malignant melanoma is heavily dependent on the stage at which it is detected.Early thin lesions are almost entirely curable.However, the 5-year survival for patients with melanomas that have spread to regional lymph nodes is 54% and for those with metastatic disease, only 13%.Therefore, understanding the determinants of late-stage melanoma diagnosis is an important cancer-control objective.Factors predictive of later stage at diagnosis and poor prognosis in previous studies are lower socioeconomic status,male sex,older age,nonwhite race,and smoking.In a previous study,we reported the effects of insurance payer and race or ethnicity on cancer stage at diagnosis. For patients diagnosed as having melanoma, we found that Medicaid patients had 4.7 times the odds of having a late-stage diagnosis and that uninsured patients had 2.6 times the odds of having a late-stage diagnosis compared with patients with commercial indemnity insurance. No racial or ethnic differences in stage at diagnosis were identified. In this study, we explored whether other sociodemographic characteristics were predictive of late-stage diagnosis. We hypothesized that a late-stage melanoma diagnosis would be associated with older age, male sex, unmarried status, lower educational attainment and income level, rural residence, and cigarette smoking.SUBJECTS AND METHODSSOURCES OF DATAThe study population consisted of all incident cases of melanoma occurring in Florida from January 1, 1994, to January 1, 1995, whose stage at diagnosis was available (N=1884). Incident cases were identified from the Florida Cancer Data System (FCDS), Florida's population-based statewide cancer registry. The FCDS was created in 1978 and has been collecting cancer incidence data since 1981. It is a member of the North American Association of Central Cancer Registries. The FCDS has well-established methods to find all cases, including cooperative arrangements with other state tumor registries and standardized procedures for quality control. The North American Association of Central Cancer Registries audits estimated that the completeness of case ascertainment for all cancers during 1990-1994 was 97% (Lydia Voti, oral communication, January 11, 2000).The FCDS is primarily a hospital-based reporting system. Hospital sources are discharge diagnostic codes in medical records, hospital pathology reports (inpatient and outpatient), radiation therapy departments, and outpatient departments (day surgery and chemotherapy clinics). Cases are also reported from private laboratories and from practicing physicians; however, finding cases in these settings is probably less complete.To include information that is not available from the FCDS (insurance payer and socioeconomic status), cases were linked with state inpatient and outpatient discharge abstracts and the 1990 US Census. The state of Florida, Agency for Health Care Administration abstracts discharge data for all admissions to nonfederal acute care hospitals, licensed ambulatory surgical centers, freestanding radiation therapy centers, and diagnostic imaging centers. Abstracted data include patient identifying information (Social Security number, date of birth, and race or ethnicity), discharge diagnoses (up to 10), and insurance payer. The calendar year 1994 is the most recent year for which all the relevant data from both the FCDS and Agency for Health Care Administration are available.The FCDS cases were matched with the Agency for Health Care Administration inpatient and outpatient discharge abstracts by a probabilistic method using all shared identifying variables (Social Security number, sex, race or ethnicity, and date of birth). Cases that successfully matched on all variables were considered valid matches. Cases were also considered valid matches if the sole discrepancy was a Social Security number or date of birth that differed by only 1 digit. Using this method, 1425 staged cases (75.6%) were successfully matched, a rate comparable to that of similar studies.For the remaining 459 cases, we relied solely on information found in the FCDS.The FCDS cases were next matched with 1990 US Census data to obtain census-derived measures of socioeconomic status. For 95.3% of cases we were able to determine the census tract based on the street address. For the remaining 4.7% of cases, we relied on the patient's ZIP code. Therefore, patients were assigned the median income and education level of either the census tract or ZIP code of their residence. We also used US Census data to determine if the patient's place of residence was urban or rural. Patients were classified as having an urban residence if their ZIP code was designated as entirely urban by the US Census. Patients were classified as having a rural residence if their ZIP code contained at least some portion that was designated as outside urban or rural by the US census. The use of census-derived measures of socioeconomic status has been validated in previous studies.VARIABLESThe main outcome examined, stage at diagnosis, was defined as the summary stage at the time of diagnosis using the SEER (Surveillance, Epidemiology, and End Results) Program's Summary Staging Guide.The summary stage is based on a combination of pathological, operative, and clinical assessments and uses all information available within 2 months of diagnosis. The stage at diagnosis is reported using categories similar to those of the SEER Program: in situ, local, regional (direct extension of disease beyond organ of origin or spread to regional nodes), or distant. For purposes of analysis, the stage at diagnosis was reclassified as either early stage (in situ or local) or late stage (regional or distant).The stage at diagnosis was available for 1884 (93.6%) of the 2012 incident cases of melanoma occurring in Florida in 1994. Patients who did not undergo staging were older (P=.01) and had lower education (P=.001) and income (P<.001) levels than patients who underwent staging (Wilcoxon rank sum test). There were no differences in race or ethnicity (P=.63), marital status (P=.09), or sex (P=.08) between staged and unstaged cases (χ2test or Wilcoxon rank sum test).The following variables were assessed as possible predictors of late-stage melanoma diagnosis: age, sex, marital status (currently married or unmarried), smoking status (current smoker or nonsmoker), education level (high school graduate or less or more than high school), household income (5 levels: <$15,000; $15,000-$24,999; $25,000-$34,999; $35,000-$49,999; and ≥$50,000), and urban vs rural residence.STATISTICAL ANALYSISRelationships between categorical variables and stage at diagnosis were assessed with the χ2test. The comparison of age differences and stage at diagnosis was assessed with the ttest. The multivariate relationship between demographic predictors and the odds of late-stage diagnosis was examined using multiple logistic regression. Logistic models included age (as a continuous variable) and indicator variables for marital status, smoking status, education level, income level, and urban vs rural residence. The patient's insurance payer and race or ethnicity were also included in all logistic models so that reported odds ratios were adjusted for these characteristics as well. In addition to their main effects, we also examined interactions between statistically significant variables in subsequent logistic models. Because the malignant potential of some in situ melanomas is uncertain, we also repeated logistic models using only the data from patients with invasive disease.Adjusted odds ratios and 95% confidence intervals are reported for each predictor variable. The statistical significance of predictor variables was assessed using the χ2likelihood ratio test.All reported Pvalues are 2-tailed. Statistical significance was set at α=.05.RESULTSThe patient characteristics are presented in Table 1. The mean (SD) age of the patients was 62.4 (16.6) years, with a range of 13 to 99 years. Most patients with melanoma were male. There were 243 patients (12.9%) diagnosed as having melanoma that had metastasized to either regional lymph nodes or distant sites. Table 2describes bivariate predictors of late-stage melanoma diagnosis. Late-stage diagnosis was more common among patients who were male, unmarried, and smokers and resided in communities with low median income and education levels. There was a statistically insignificant trend for patients residing in rural communities to be diagnosed at later stage than patients from urban settings. Patients diagnosed as having late-stage disease were similar in age to those diagnosed as having early-stage disease (60.6 vs 62.3 years; ttest, 1.51; P=.13).Table 1. Characteristics of 1884 Patients Diagnosed as Having Melanoma in Florida, 1994&ast;CharacteristicNo. (%) of PatientsAge, y<3072 (3.8)30-49391 (20.7)50-69676 (35.9)≥70745 (39.5)Education level, median†High school or less659 (35.0)More than high school1225 (65.0)Male1117 (59.3)Female767 (40.7)Household income, median, $<15 00042 (2.2)15 000-24 999440 (23.4)25 000-34 999973 (51.6)35 000-49 999348 (18.5)≥50 00072 (3.8)Insurance payerMedicare669 (35.5)Medicaid21 (1.1)Medicare and Medicaid HMO71 (3.8)Commercial HMO180 (9.6)Commercial indemnity274 (14.5)Commercial PPO186 (9.9)Uninsured72 (3.8)Other58 (3.1)Married1268 (67.3)Unmarried525 (27.9)Urban residence956 (50.7)Rural residence928 (49.3)White1763 (93.6)Nonwhite121 (6.4)Nonsmoker1632 (86.6)Smoker252 (13.4)Stage at diagnosisIn situ295 (15.7)Local1346 (71.4)Regional129 (6.8)Distant114 (6.1)&ast;Information on education level and household income was obtained from the census tract or ZIP code of residence. HMO indicates health maintenance organization; and PPO, preferred provider organization.Table 2. Predictors of Late-Stage Diagnosis of Melanoma&ast;VariableNo. (%) of Patients With Late-Stage Diagnoses (N=1884)PEducation levelHigh school or less115/659 (17.5)<.001More than high school128/1225 (10.4)Male170/1117 (15.2)<.001Female73/767 (9.5)Income level, $<15 0006/42 (14.3)15 000-24 99978/440 (17.7)25 000-34 999118/973 (12.1)<.00135 000-49 99932/348 (9.2)≥50 0007/72 (9.7)Married151/1268 (11.9).01Unmarried86/525 (16.4)Urban residence112/956 (11.7).12Rural residence131/928 (14.1)Nonsmoker181/1632 (11.1)<.001Smoker62/252 (24.6)&ast;Late stage was defined as regional or distant metastases. Pvalues were determined using the χ2test. Information on education and income levels was determined by census tract or ZIP code of residence. Information on income level was unavailable in 9 patients (2 of whom had late-stage diagnoses). Information on marital status was unavailable in 91 patients (6 of whom had late-stage diagnoses).Multivariate predictors of late-stage melanoma diagnosis are presented in Table 3. Patients who were unmarried, male, and smokers and resided in communities with a median educational attainment of high school or less were more likely to be diagnosed as having melanoma at a late stage. The stage at diagnosis was unrelated to the patient's age, community measures of median income, or urban vs rural residence. There were no statistically significant interactions between sex, marital status, smoking status, or education level in the logistic regression models. The results of logistic regression analysis were also unchanged when in situ cases were excluded from analysis.Table 3. Multivariate Predictors of Late-Stage Melanoma Diagnosis in 1470 Patients&ast;PredictorOdds Ratio (95% Confidence Interval)PAge at diagnosis†1.01 (0.99-1.02).46Education levelMore than high school1.00 (Reference). . .High school or less1.45 (1.00-2.12).048Female1.00 (Reference). . .Male2.18 (1.54-3.09)<.001Income level, median†0.92 (0.73-1.12).46Married1.00 (Reference). . .Unmarried1.53 (1.08-2.16).01Rural residence1.00 (Reference). . .Urban residence0.93 (0.68-1.27).63Nonsmoker1.00 (Reference). . .Smoker2.20 (1.51-3.20)<.001&ast;Odds ratios for late-stage diagnosis (regional or distant metastases) were simultaneously adjusted for age, sex, race, marital status, education level, income level, smoking status, insurance payer, and urban vs rural residence using multiple logistic regression. Sample excludes 414 patients with missing data. Ellipses indicate not applicable.†Odds ratios represent changes in odds of late-stage diagnosis per year of increasing age or per unit of increasing income level.COMMENTWe found that the following sociodemographic groups were more likely to be diagnosed as having melanoma at a late stage: male patients, unmarried patients, smokers, and those who resided in communities with low median educational attainment. Among these factors, male sex and smoking had the most pronounced effects, with more than double the odds of late-stage diagnosis.Our finding that males were more likely to have melanoma diagnosed at a late stage agrees with findings in other studies.Men are more likely than women to present with nodular melanomas and melanomas located on the back.Tumors in less visible body areas are significantly thicker at the time of diagnosis than those occurring in more highly visible areas, most likely as a result of delayed detection.Consistent with this view, a study by Koh et alfound that men were less likely than women to identify their own melanoma. In addition, nodular melanomas are less likely to be detected during visits to physicians for unrelated problems and instead are discovered more often by patients when lesions become symptomatic.Men are also less likely to report undergoing either self-screening or physician screening.In addition, men are also less knowledgeable about melanoma and have less favorable attitudes.The finding that unmarried persons were more likely to be diagnosed as having late-stage melanoma is also consistent with the finding that patients had difficulty detecting less visible lesions on self-examination. Studies by Koh et alconfirmed the importance of a spouse in detecting many melanomas. In their study, a spouse was the third most common person (after the index patient and a physician) to detect a melanoma. This finding suggests that physician surveillance of melanoma is especially important for unmarried patients.Similar to the findings in other studies, we found that smokers were more likely to be diagnosed as having melanoma at late stage.Among the variables studied, smoking was the strongest predictor of a diagnosis of melanoma at a late stage. It has been postulated that smoking leads to a diminished host-immune response against melanomas, leading to greater tumor virulence. It is also possible that smoking is a marker for less active skin surveillance, which would also lead to later detection of melanomas.We found that patients residing in areas with a low median educational attainment were more likely to have a late-stage diagnosis. Otherssimilarly found that patients from more affluent areas generally have thinner melanomas at the time of detection. Moreover, patients with melanoma with a lower socioeconomic status are more likely to die of their disease.Patients with a lower socioeconomic status are less likely to undergo either self-screening or physician screening.Lower socioeconomic status has also been associated with less knowledge and awareness of melanoma.In the results of multivariate analysis, we found an effect for education level, but not for income.We found no significant linear or nonlinear relationship between age and stage at diagnosis. Most studieshave concluded that elderly patients are more likely to be diagnosed at a late stage. The higher frequency of thick melanomas in older patients has been attributed to an increased proportion of nodular melanomas and a decreased ability to recognize the changes of melanoma.Our finding that there was no relationship between age and stage at diagnosis may be peculiar to Florida, with its large number of elderly retirees from other states. One would expect that elderly persons who are able to retire to another state are generally healthier, better educated, and better off financially than those who are not able to retire out of state. Miller et alfound that elderly patients are more likely to do self-examinations.This study has a number of important limitations. Initially, socioeconomic status was measured at a community not an individual level. However, previous studieshave validated the use of aggregate measures of socioeconomic status. We also were limited to the SEER Program's categories of staging (in situ, local, regional, and distant) rather than more detailed staging measures, such as Breslow thickness or Clark level. As cancers are detected and managed increasingly in the outpatient setting, it will become more difficult for tumor registries to find cases. There are no well-established mechanisms for gathering the detailed information in the FCDS about outpatients treated in physicians' offices. Late-stage melanomas are probably more likely than early-stage lesions to be reported to tumor registries.An additional concern of skin cancer screening is that greater surveillance could lead to the detection of in situ melanomas with uncertain malignant potential.Our findings were similar, however, when we analyzed only cases of invasive melanomas. Finally, our study was restricted to incident cases of malignant melanoma in Florida, which may not be representative of other diseases or other parts of the country.In conclusion, we found that males, unmarried patients, smokers, and those who resided in communities with low median educational attainment were more likely to be diagnosed as having melanoma at a late stage. There are several implications of these findings. Although most physicians know that family history and excess sun exposure are risk factors for melanoma, they should be aware of the other factors that may increase the likelihood of late-stage cancer and should screen patients meeting these criteria more conscientiously. Primary care physicians should screen patients with a higher risk whenever they are in the office, even if the complaints are unrelated to melanoma. More physicians have been trained to recognize risk factors for cardiovascular disease. The same attention should be focused on melanoma detection. Many physicians use problem lists and flow sheets on their office charts. These charts could include an indication as to whether the skin was examined fully and whether self-examination information was given to patients with a higher risk of melanoma.The American Cancer Society and the American Academy of Dermatology have launched public education efforts to increase awareness of skin cancer. These efforts should be targeted to patients with a higher risk using appropriate messages and the media. Information should recommend increased self-surveillance and checking the partner's skin.BMillerLRiesBHankeyCKosaryBEdwardsCancer Statistics Review: 1973-1989.Bethesda, Md: US Dept of Health and Human Services; 1992. NIH publication 92-278.RJFriedmanDSRigelMKSilvermanAWKopfKAVossaertMalignant melanoma in the 1990s: the continued importance of early detection and the role of physician examination and self-examination of the skin.CA Cancer J Clin.1991;41:201-226.SLandisTMurraySBoldenPWingoCancer statistics, 1998.CA Cancer J Clin.1998;48:6-29.DRMillerACGellerSWWyattMelanoma awareness and self-examination practices: results of a United States survey.J Am Acad Dermatol.1996;34:962-970.IAhmedMalignant melanoma: prognostic indicators.Mayo Clin Proc.1997;72:356-361.AGirgisEMCampbellSRedmanRWSanson-FisherScreening for melanoma: a community survey of prevalence and predictors.Med J Aust.1991;154:338-343.ACGellerDRMillerRALewRWClappMBWennekerHKKohCutaneous melanoma mortality among the socioeconomically disadvantaged in Massachusetts.Am J Public Health.1996;86:538-544.AEChangLHKarnellHRMenckfor the American College of Surgeons Commission on Cancer and the American Cancer SocietyThe National Cancer Data Base report on cutaneous and noncutaneous melanoma: a summary of 84,836 cases from the past decade.Cancer.1998;83:1664-1678.ABonettDRoderAEstermanEpidemiological features of melanoma in South Australia: implications for cancer control.Med J Aust.1989;151:502-504, 506-509.DMRoderCGLukeKAMcCaulAJEstermanTrends in prognostic factors of melanoma in South Australia, 1981-1992: implications for health promotion.Med J Aust.1995;162:25-29.DVageroGPerssonRisks, survival and trends of malignant melanoma among white and blue collar workers in Sweden.Soc Sci Med.1984;19:475-478.RDCressEAHollyIncidence of cutaneous melanoma among non-Hispanic whites, Hispanics, Asians, and blacks: an analysis of California cancer registry data, 1988-93.Cancer Causes Control.1997;8:246-252.PHerseyRWSillarCGHoweFactors related to the presentation of patients with thick primary melanomas.Med J Aust.1991;154:583-587.PFHanrahanPHerseyCAD'EsteFactors involved in presentation of older people with thick melanoma.Med J Aust.1998;169:410-414.MThornFPontenRBergstromPSparenHOAdamiClinical and histopathologic predictors of survival in patients with malignant melanoma: a population-based study in Sweden.J Natl Cancer Inst.1994;86:761-769.PFAustinCWCruseGLymanKSchroerFGlassDSReintgenAge as a prognostic factor in the malignant melanoma population.Ann Surg Oncol.1994;1:487-494.JSGoodwinJMSametCRKeyCHumbleDKutvirtCHuntStage at diagnosis of cancer varies with the age of the patient.J Am Geriatr Soc.1986;34:20-26.PFHanrahanPHerseyABWatsonTMCallaghanThe effect of an educational brochure on knowledge and early detection of melanoma.Aust J Public Health.1995;19:270-274.FHolmesEHearneCancer stage-to-age relationship: implications for cancer screening in the elderly.J Am Geriatr Soc.1981;29:55-57.AVayerATLeforCutaneous melanoma in African-Americans.South Med J.1993;86:181-182.HKohASoberCJDayRLewTFitzpatrickCigarette smoking and malignant melanoma: prognostic implications.Cancer.1984;53:2570-2573.DRigelRFriedmanJLevineAKopfMLevensteinCigarette smoking and malignant melanoma: prognostic implications.J Dermatol Surg Oncol.1981;7:889-891.HShawGMiltonWMcCarthyGFaragoPDilworthEffect of smoking on the recurrence of malignant melanoma.Med J Aust.1979;1:208-209.HShawGMiltonSmoking and the development of metastases from malignant melanoma.Int J Cancer.1981;28:153-156.RRoetzheimNPalCTennantEffects of health insurance and race on the early detection of cancer.J Natl Cancer Inst.1999;91:1409-1415.JAyanianBKohlerTAbeAEpsteinThe relation between health insurance coverage and clinical outcomes among women with breast cancer.N Engl J Med.1993;329:326-331.THoferRWolfePTedeschiLMcMahonJGriffithUse of community versus individual socioeconomic data predicting variation in hospital use.Health Serv Res.1998;33:243-259.NKriegerOvercoming the absence of socioeconomic data in medical records: validation and application of a census-based methodology.Am J Public Health.1992;82:703-710.NKriegerEFeeSocial class: the missing link in U.S. health data.Int J Health Serv.1994;24:25-44.EShambaughMWeissSummary Staging Guide: Cancer Surveillance Epidemiology and End Results Reporting: SEER Program.Bethesda, Md: National Cancer Institute; 1977.DHosmerSLemeshowApplied Logistic Regression.New York, NY: John Wiley & Sons; 1989.LTemoshokRJDiClementeDMSweetMSBloisRWSagebielFactors related to patient delay in seeking medical attention for cutaneous malignant melanoma.Cancer.1984;54:3048-3053.YHemoMGutmanJMKlausnerAnatomic site of primary melanoma is associated with depth of invasion.Arch Surg.1999;134:148-150.LTemoshokBiopsychosocial studies on cutaneous malignant melanoma: psychosocial factors associated with prognostic indicators, progression, psychophysiology and tumor-host response.Soc Sci Med.1985;20:833-840.HKKohDRMillerACGellerRWClappMBMercerRALewWho discovers melanoma? patterns from a population-based survey.J Am Acad Dermatol.1992;26:914-919.HKKohACGellerDRMillerACarusoIGageRALewWho is being screened for melanoma/skin cancer? characteristics of persons screened in Massachusetts.J Am Acad Dermatol.1991;24:271-277.YBrandbergCBolundHMichelsonEMansson-BrahmeURingborgPOSjodenPerceived susceptibility to and knowledge of malignant melanoma: screening participants vs the general population.Prev Med.1996;25:170-177.ADiez-RouxBringing context back into epidemiology: variables and fallacies in multilevel analysis.Am J Public Health.1998;88:216-222.JMeliaTFrostRGraham-BrownProblems with registration of cutaneous malignant melanoma in England.Br J Cancer.1995;72:224-228.RASwerlickSChenThe melanoma epidemic: is increased surveillance the solution or the problem?Arch Dermatol.1996;132:881-884.RASwerlickSChenThe melanoma epidemic: more apparent than real?Mayo Clin Proc.1997;72:559-564.Accepted for publication October 22, 1999.Richard G. Roetzheim, MD, MSPH, was supported by a Generalist Physician Faculty Scholars Award from the Robert Wood Johnson Foundation, Princeton, NJ.Corresponding author and reprints: Daniel J. Van Durme, MD, University of South Florida, Department of Family Medicine, 12901 Bruce B. Downs Blvd, Box 13, Tampa, FL 33612 (e-mail: dvandurm@com1.med.usf.edu). http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Archives of Family Medicine American Medical Association

Demographic Predictors of Melanoma Stage at Diagnosis

Download PDF
6 pages

Loading next page...
 
/lp/american-medical-association/demographic-predictors-of-melanoma-stage-at-diagnosis-Vu2TOmX1Ef
Publisher
American Medical Association
Copyright
Copyright 2000 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
ISSN
1063-3987
eISSN
1063-3987
DOI
10-1001/pubs.Arch Fam Med.-ISSN-1063-3987-9-7-foc9078
Publisher site

Abstract

ObjectiveTo examine sociodemographic characteristics as possible predictors of late-stage melanoma diagnosis. We hypothesized that late-stage diagnosis would be associated with the following: older age, male sex, unmarried status, lower educational attainment and income level, rural residence, and cigarette smoking.MethodsWe used data from the state tumor registry to study all incident cases of melanoma occurring in Florida during 1994 whose stage at diagnosis was available (N=1884). We used multiple logistic regression to determine the effects of sociodemographic characteristics on the odds of late-stage (regional or distant metastases) diagnosis.ResultsThere were 243 patients (12.9%) diagnosed as having melanoma that had metastasized to either regional lymph nodes or distant sites. Patients who were unmarried (odds ratio, 1.5; P=.01), male (odds ratio, 2.2; P<.001), or smokers (odds ratio, 2.2; P<.001) or who resided in communities with lower median educational attainment (odds ratio, 1.5; P=.048) had greater odds of having a late-stage diagnosis.ConclusionsTo detect these cancers at an earlier stage and improve outcomes, there should be increased educational efforts directed toward physicians who treat these patients. A recognition that there may be additional risk factors for late-stage diagnosis, beyond the established risk factors, such as family history and excess sun exposure, should be included in the initial assessment. Specific public education efforts should also be targeted to these patients to increase their self-surveillance and surveillance of their partners.DURING 1973-1989, the incidence of malignant melanoma increased 85%, more than any other major cancer.There were approximately 42,000 new cases in 1998, with 7000 deaths.The lifetime risk of acquiring melanoma is now estimated to be about 1 in 87.Within the next 10 years, the risk is expected to climb to 1 in 75.The prognosis of malignant melanoma is heavily dependent on the stage at which it is detected.Early thin lesions are almost entirely curable.However, the 5-year survival for patients with melanomas that have spread to regional lymph nodes is 54% and for those with metastatic disease, only 13%.Therefore, understanding the determinants of late-stage melanoma diagnosis is an important cancer-control objective.Factors predictive of later stage at diagnosis and poor prognosis in previous studies are lower socioeconomic status,male sex,older age,nonwhite race,and smoking.In a previous study,we reported the effects of insurance payer and race or ethnicity on cancer stage at diagnosis. For patients diagnosed as having melanoma, we found that Medicaid patients had 4.7 times the odds of having a late-stage diagnosis and that uninsured patients had 2.6 times the odds of having a late-stage diagnosis compared with patients with commercial indemnity insurance. No racial or ethnic differences in stage at diagnosis were identified. In this study, we explored whether other sociodemographic characteristics were predictive of late-stage diagnosis. We hypothesized that a late-stage melanoma diagnosis would be associated with older age, male sex, unmarried status, lower educational attainment and income level, rural residence, and cigarette smoking.SUBJECTS AND METHODSSOURCES OF DATAThe study population consisted of all incident cases of melanoma occurring in Florida from January 1, 1994, to January 1, 1995, whose stage at diagnosis was available (N=1884). Incident cases were identified from the Florida Cancer Data System (FCDS), Florida's population-based statewide cancer registry. The FCDS was created in 1978 and has been collecting cancer incidence data since 1981. It is a member of the North American Association of Central Cancer Registries. The FCDS has well-established methods to find all cases, including cooperative arrangements with other state tumor registries and standardized procedures for quality control. The North American Association of Central Cancer Registries audits estimated that the completeness of case ascertainment for all cancers during 1990-1994 was 97% (Lydia Voti, oral communication, January 11, 2000).The FCDS is primarily a hospital-based reporting system. Hospital sources are discharge diagnostic codes in medical records, hospital pathology reports (inpatient and outpatient), radiation therapy departments, and outpatient departments (day surgery and chemotherapy clinics). Cases are also reported from private laboratories and from practicing physicians; however, finding cases in these settings is probably less complete.To include information that is not available from the FCDS (insurance payer and socioeconomic status), cases were linked with state inpatient and outpatient discharge abstracts and the 1990 US Census. The state of Florida, Agency for Health Care Administration abstracts discharge data for all admissions to nonfederal acute care hospitals, licensed ambulatory surgical centers, freestanding radiation therapy centers, and diagnostic imaging centers. Abstracted data include patient identifying information (Social Security number, date of birth, and race or ethnicity), discharge diagnoses (up to 10), and insurance payer. The calendar year 1994 is the most recent year for which all the relevant data from both the FCDS and Agency for Health Care Administration are available.The FCDS cases were matched with the Agency for Health Care Administration inpatient and outpatient discharge abstracts by a probabilistic method using all shared identifying variables (Social Security number, sex, race or ethnicity, and date of birth). Cases that successfully matched on all variables were considered valid matches. Cases were also considered valid matches if the sole discrepancy was a Social Security number or date of birth that differed by only 1 digit. Using this method, 1425 staged cases (75.6%) were successfully matched, a rate comparable to that of similar studies.For the remaining 459 cases, we relied solely on information found in the FCDS.The FCDS cases were next matched with 1990 US Census data to obtain census-derived measures of socioeconomic status. For 95.3% of cases we were able to determine the census tract based on the street address. For the remaining 4.7% of cases, we relied on the patient's ZIP code. Therefore, patients were assigned the median income and education level of either the census tract or ZIP code of their residence. We also used US Census data to determine if the patient's place of residence was urban or rural. Patients were classified as having an urban residence if their ZIP code was designated as entirely urban by the US Census. Patients were classified as having a rural residence if their ZIP code contained at least some portion that was designated as outside urban or rural by the US census. The use of census-derived measures of socioeconomic status has been validated in previous studies.VARIABLESThe main outcome examined, stage at diagnosis, was defined as the summary stage at the time of diagnosis using the SEER (Surveillance, Epidemiology, and End Results) Program's Summary Staging Guide.The summary stage is based on a combination of pathological, operative, and clinical assessments and uses all information available within 2 months of diagnosis. The stage at diagnosis is reported using categories similar to those of the SEER Program: in situ, local, regional (direct extension of disease beyond organ of origin or spread to regional nodes), or distant. For purposes of analysis, the stage at diagnosis was reclassified as either early stage (in situ or local) or late stage (regional or distant).The stage at diagnosis was available for 1884 (93.6%) of the 2012 incident cases of melanoma occurring in Florida in 1994. Patients who did not undergo staging were older (P=.01) and had lower education (P=.001) and income (P<.001) levels than patients who underwent staging (Wilcoxon rank sum test). There were no differences in race or ethnicity (P=.63), marital status (P=.09), or sex (P=.08) between staged and unstaged cases (χ2test or Wilcoxon rank sum test).The following variables were assessed as possible predictors of late-stage melanoma diagnosis: age, sex, marital status (currently married or unmarried), smoking status (current smoker or nonsmoker), education level (high school graduate or less or more than high school), household income (5 levels: <$15,000; $15,000-$24,999; $25,000-$34,999; $35,000-$49,999; and ≥$50,000), and urban vs rural residence.STATISTICAL ANALYSISRelationships between categorical variables and stage at diagnosis were assessed with the χ2test. The comparison of age differences and stage at diagnosis was assessed with the ttest. The multivariate relationship between demographic predictors and the odds of late-stage diagnosis was examined using multiple logistic regression. Logistic models included age (as a continuous variable) and indicator variables for marital status, smoking status, education level, income level, and urban vs rural residence. The patient's insurance payer and race or ethnicity were also included in all logistic models so that reported odds ratios were adjusted for these characteristics as well. In addition to their main effects, we also examined interactions between statistically significant variables in subsequent logistic models. Because the malignant potential of some in situ melanomas is uncertain, we also repeated logistic models using only the data from patients with invasive disease.Adjusted odds ratios and 95% confidence intervals are reported for each predictor variable. The statistical significance of predictor variables was assessed using the χ2likelihood ratio test.All reported Pvalues are 2-tailed. Statistical significance was set at α=.05.RESULTSThe patient characteristics are presented in Table 1. The mean (SD) age of the patients was 62.4 (16.6) years, with a range of 13 to 99 years. Most patients with melanoma were male. There were 243 patients (12.9%) diagnosed as having melanoma that had metastasized to either regional lymph nodes or distant sites. Table 2describes bivariate predictors of late-stage melanoma diagnosis. Late-stage diagnosis was more common among patients who were male, unmarried, and smokers and resided in communities with low median income and education levels. There was a statistically insignificant trend for patients residing in rural communities to be diagnosed at later stage than patients from urban settings. Patients diagnosed as having late-stage disease were similar in age to those diagnosed as having early-stage disease (60.6 vs 62.3 years; ttest, 1.51; P=.13).Table 1. Characteristics of 1884 Patients Diagnosed as Having Melanoma in Florida, 1994&ast;CharacteristicNo. (%) of PatientsAge, y<3072 (3.8)30-49391 (20.7)50-69676 (35.9)≥70745 (39.5)Education level, median†High school or less659 (35.0)More than high school1225 (65.0)Male1117 (59.3)Female767 (40.7)Household income, median, $<15 00042 (2.2)15 000-24 999440 (23.4)25 000-34 999973 (51.6)35 000-49 999348 (18.5)≥50 00072 (3.8)Insurance payerMedicare669 (35.5)Medicaid21 (1.1)Medicare and Medicaid HMO71 (3.8)Commercial HMO180 (9.6)Commercial indemnity274 (14.5)Commercial PPO186 (9.9)Uninsured72 (3.8)Other58 (3.1)Married1268 (67.3)Unmarried525 (27.9)Urban residence956 (50.7)Rural residence928 (49.3)White1763 (93.6)Nonwhite121 (6.4)Nonsmoker1632 (86.6)Smoker252 (13.4)Stage at diagnosisIn situ295 (15.7)Local1346 (71.4)Regional129 (6.8)Distant114 (6.1)&ast;Information on education level and household income was obtained from the census tract or ZIP code of residence. HMO indicates health maintenance organization; and PPO, preferred provider organization.Table 2. Predictors of Late-Stage Diagnosis of Melanoma&ast;VariableNo. (%) of Patients With Late-Stage Diagnoses (N=1884)PEducation levelHigh school or less115/659 (17.5)<.001More than high school128/1225 (10.4)Male170/1117 (15.2)<.001Female73/767 (9.5)Income level, $<15 0006/42 (14.3)15 000-24 99978/440 (17.7)25 000-34 999118/973 (12.1)<.00135 000-49 99932/348 (9.2)≥50 0007/72 (9.7)Married151/1268 (11.9).01Unmarried86/525 (16.4)Urban residence112/956 (11.7).12Rural residence131/928 (14.1)Nonsmoker181/1632 (11.1)<.001Smoker62/252 (24.6)&ast;Late stage was defined as regional or distant metastases. Pvalues were determined using the χ2test. Information on education and income levels was determined by census tract or ZIP code of residence. Information on income level was unavailable in 9 patients (2 of whom had late-stage diagnoses). Information on marital status was unavailable in 91 patients (6 of whom had late-stage diagnoses).Multivariate predictors of late-stage melanoma diagnosis are presented in Table 3. Patients who were unmarried, male, and smokers and resided in communities with a median educational attainment of high school or less were more likely to be diagnosed as having melanoma at a late stage. The stage at diagnosis was unrelated to the patient's age, community measures of median income, or urban vs rural residence. There were no statistically significant interactions between sex, marital status, smoking status, or education level in the logistic regression models. The results of logistic regression analysis were also unchanged when in situ cases were excluded from analysis.Table 3. Multivariate Predictors of Late-Stage Melanoma Diagnosis in 1470 Patients&ast;PredictorOdds Ratio (95% Confidence Interval)PAge at diagnosis†1.01 (0.99-1.02).46Education levelMore than high school1.00 (Reference). . .High school or less1.45 (1.00-2.12).048Female1.00 (Reference). . .Male2.18 (1.54-3.09)<.001Income level, median†0.92 (0.73-1.12).46Married1.00 (Reference). . .Unmarried1.53 (1.08-2.16).01Rural residence1.00 (Reference). . .Urban residence0.93 (0.68-1.27).63Nonsmoker1.00 (Reference). . .Smoker2.20 (1.51-3.20)<.001&ast;Odds ratios for late-stage diagnosis (regional or distant metastases) were simultaneously adjusted for age, sex, race, marital status, education level, income level, smoking status, insurance payer, and urban vs rural residence using multiple logistic regression. Sample excludes 414 patients with missing data. Ellipses indicate not applicable.†Odds ratios represent changes in odds of late-stage diagnosis per year of increasing age or per unit of increasing income level.COMMENTWe found that the following sociodemographic groups were more likely to be diagnosed as having melanoma at a late stage: male patients, unmarried patients, smokers, and those who resided in communities with low median educational attainment. Among these factors, male sex and smoking had the most pronounced effects, with more than double the odds of late-stage diagnosis.Our finding that males were more likely to have melanoma diagnosed at a late stage agrees with findings in other studies.Men are more likely than women to present with nodular melanomas and melanomas located on the back.Tumors in less visible body areas are significantly thicker at the time of diagnosis than those occurring in more highly visible areas, most likely as a result of delayed detection.Consistent with this view, a study by Koh et alfound that men were less likely than women to identify their own melanoma. In addition, nodular melanomas are less likely to be detected during visits to physicians for unrelated problems and instead are discovered more often by patients when lesions become symptomatic.Men are also less likely to report undergoing either self-screening or physician screening.In addition, men are also less knowledgeable about melanoma and have less favorable attitudes.The finding that unmarried persons were more likely to be diagnosed as having late-stage melanoma is also consistent with the finding that patients had difficulty detecting less visible lesions on self-examination. Studies by Koh et alconfirmed the importance of a spouse in detecting many melanomas. In their study, a spouse was the third most common person (after the index patient and a physician) to detect a melanoma. This finding suggests that physician surveillance of melanoma is especially important for unmarried patients.Similar to the findings in other studies, we found that smokers were more likely to be diagnosed as having melanoma at late stage.Among the variables studied, smoking was the strongest predictor of a diagnosis of melanoma at a late stage. It has been postulated that smoking leads to a diminished host-immune response against melanomas, leading to greater tumor virulence. It is also possible that smoking is a marker for less active skin surveillance, which would also lead to later detection of melanomas.We found that patients residing in areas with a low median educational attainment were more likely to have a late-stage diagnosis. Otherssimilarly found that patients from more affluent areas generally have thinner melanomas at the time of detection. Moreover, patients with melanoma with a lower socioeconomic status are more likely to die of their disease.Patients with a lower socioeconomic status are less likely to undergo either self-screening or physician screening.Lower socioeconomic status has also been associated with less knowledge and awareness of melanoma.In the results of multivariate analysis, we found an effect for education level, but not for income.We found no significant linear or nonlinear relationship between age and stage at diagnosis. Most studieshave concluded that elderly patients are more likely to be diagnosed at a late stage. The higher frequency of thick melanomas in older patients has been attributed to an increased proportion of nodular melanomas and a decreased ability to recognize the changes of melanoma.Our finding that there was no relationship between age and stage at diagnosis may be peculiar to Florida, with its large number of elderly retirees from other states. One would expect that elderly persons who are able to retire to another state are generally healthier, better educated, and better off financially than those who are not able to retire out of state. Miller et alfound that elderly patients are more likely to do self-examinations.This study has a number of important limitations. Initially, socioeconomic status was measured at a community not an individual level. However, previous studieshave validated the use of aggregate measures of socioeconomic status. We also were limited to the SEER Program's categories of staging (in situ, local, regional, and distant) rather than more detailed staging measures, such as Breslow thickness or Clark level. As cancers are detected and managed increasingly in the outpatient setting, it will become more difficult for tumor registries to find cases. There are no well-established mechanisms for gathering the detailed information in the FCDS about outpatients treated in physicians' offices. Late-stage melanomas are probably more likely than early-stage lesions to be reported to tumor registries.An additional concern of skin cancer screening is that greater surveillance could lead to the detection of in situ melanomas with uncertain malignant potential.Our findings were similar, however, when we analyzed only cases of invasive melanomas. Finally, our study was restricted to incident cases of malignant melanoma in Florida, which may not be representative of other diseases or other parts of the country.In conclusion, we found that males, unmarried patients, smokers, and those who resided in communities with low median educational attainment were more likely to be diagnosed as having melanoma at a late stage. There are several implications of these findings. Although most physicians know that family history and excess sun exposure are risk factors for melanoma, they should be aware of the other factors that may increase the likelihood of late-stage cancer and should screen patients meeting these criteria more conscientiously. Primary care physicians should screen patients with a higher risk whenever they are in the office, even if the complaints are unrelated to melanoma. More physicians have been trained to recognize risk factors for cardiovascular disease. The same attention should be focused on melanoma detection. Many physicians use problem lists and flow sheets on their office charts. These charts could include an indication as to whether the skin was examined fully and whether self-examination information was given to patients with a higher risk of melanoma.The American Cancer Society and the American Academy of Dermatology have launched public education efforts to increase awareness of skin cancer. These efforts should be targeted to patients with a higher risk using appropriate messages and the media. Information should recommend increased self-surveillance and checking the partner's skin.BMillerLRiesBHankeyCKosaryBEdwardsCancer Statistics Review: 1973-1989.Bethesda, Md: US Dept of Health and Human Services; 1992. NIH publication 92-278.RJFriedmanDSRigelMKSilvermanAWKopfKAVossaertMalignant melanoma in the 1990s: the continued importance of early detection and the role of physician examination and self-examination of the skin.CA Cancer J Clin.1991;41:201-226.SLandisTMurraySBoldenPWingoCancer statistics, 1998.CA Cancer J Clin.1998;48:6-29.DRMillerACGellerSWWyattMelanoma awareness and self-examination practices: results of a United States survey.J Am Acad Dermatol.1996;34:962-970.IAhmedMalignant melanoma: prognostic indicators.Mayo Clin Proc.1997;72:356-361.AGirgisEMCampbellSRedmanRWSanson-FisherScreening for melanoma: a community survey of prevalence and predictors.Med J Aust.1991;154:338-343.ACGellerDRMillerRALewRWClappMBWennekerHKKohCutaneous melanoma mortality among the socioeconomically disadvantaged in Massachusetts.Am J Public Health.1996;86:538-544.AEChangLHKarnellHRMenckfor the American College of Surgeons Commission on Cancer and the American Cancer SocietyThe National Cancer Data Base report on cutaneous and noncutaneous melanoma: a summary of 84,836 cases from the past decade.Cancer.1998;83:1664-1678.ABonettDRoderAEstermanEpidemiological features of melanoma in South Australia: implications for cancer control.Med J Aust.1989;151:502-504, 506-509.DMRoderCGLukeKAMcCaulAJEstermanTrends in prognostic factors of melanoma in South Australia, 1981-1992: implications for health promotion.Med J Aust.1995;162:25-29.DVageroGPerssonRisks, survival and trends of malignant melanoma among white and blue collar workers in Sweden.Soc Sci Med.1984;19:475-478.RDCressEAHollyIncidence of cutaneous melanoma among non-Hispanic whites, Hispanics, Asians, and blacks: an analysis of California cancer registry data, 1988-93.Cancer Causes Control.1997;8:246-252.PHerseyRWSillarCGHoweFactors related to the presentation of patients with thick primary melanomas.Med J Aust.1991;154:583-587.PFHanrahanPHerseyCAD'EsteFactors involved in presentation of older people with thick melanoma.Med J Aust.1998;169:410-414.MThornFPontenRBergstromPSparenHOAdamiClinical and histopathologic predictors of survival in patients with malignant melanoma: a population-based study in Sweden.J Natl Cancer Inst.1994;86:761-769.PFAustinCWCruseGLymanKSchroerFGlassDSReintgenAge as a prognostic factor in the malignant melanoma population.Ann Surg Oncol.1994;1:487-494.JSGoodwinJMSametCRKeyCHumbleDKutvirtCHuntStage at diagnosis of cancer varies with the age of the patient.J Am Geriatr Soc.1986;34:20-26.PFHanrahanPHerseyABWatsonTMCallaghanThe effect of an educational brochure on knowledge and early detection of melanoma.Aust J Public Health.1995;19:270-274.FHolmesEHearneCancer stage-to-age relationship: implications for cancer screening in the elderly.J Am Geriatr Soc.1981;29:55-57.AVayerATLeforCutaneous melanoma in African-Americans.South Med J.1993;86:181-182.HKohASoberCJDayRLewTFitzpatrickCigarette smoking and malignant melanoma: prognostic implications.Cancer.1984;53:2570-2573.DRigelRFriedmanJLevineAKopfMLevensteinCigarette smoking and malignant melanoma: prognostic implications.J Dermatol Surg Oncol.1981;7:889-891.HShawGMiltonWMcCarthyGFaragoPDilworthEffect of smoking on the recurrence of malignant melanoma.Med J Aust.1979;1:208-209.HShawGMiltonSmoking and the development of metastases from malignant melanoma.Int J Cancer.1981;28:153-156.RRoetzheimNPalCTennantEffects of health insurance and race on the early detection of cancer.J Natl Cancer Inst.1999;91:1409-1415.JAyanianBKohlerTAbeAEpsteinThe relation between health insurance coverage and clinical outcomes among women with breast cancer.N Engl J Med.1993;329:326-331.THoferRWolfePTedeschiLMcMahonJGriffithUse of community versus individual socioeconomic data predicting variation in hospital use.Health Serv Res.1998;33:243-259.NKriegerOvercoming the absence of socioeconomic data in medical records: validation and application of a census-based methodology.Am J Public Health.1992;82:703-710.NKriegerEFeeSocial class: the missing link in U.S. health data.Int J Health Serv.1994;24:25-44.EShambaughMWeissSummary Staging Guide: Cancer Surveillance Epidemiology and End Results Reporting: SEER Program.Bethesda, Md: National Cancer Institute; 1977.DHosmerSLemeshowApplied Logistic Regression.New York, NY: John Wiley & Sons; 1989.LTemoshokRJDiClementeDMSweetMSBloisRWSagebielFactors related to patient delay in seeking medical attention for cutaneous malignant melanoma.Cancer.1984;54:3048-3053.YHemoMGutmanJMKlausnerAnatomic site of primary melanoma is associated with depth of invasion.Arch Surg.1999;134:148-150.LTemoshokBiopsychosocial studies on cutaneous malignant melanoma: psychosocial factors associated with prognostic indicators, progression, psychophysiology and tumor-host response.Soc Sci Med.1985;20:833-840.HKKohDRMillerACGellerRWClappMBMercerRALewWho discovers melanoma? patterns from a population-based survey.J Am Acad Dermatol.1992;26:914-919.HKKohACGellerDRMillerACarusoIGageRALewWho is being screened for melanoma/skin cancer? characteristics of persons screened in Massachusetts.J Am Acad Dermatol.1991;24:271-277.YBrandbergCBolundHMichelsonEMansson-BrahmeURingborgPOSjodenPerceived susceptibility to and knowledge of malignant melanoma: screening participants vs the general population.Prev Med.1996;25:170-177.ADiez-RouxBringing context back into epidemiology: variables and fallacies in multilevel analysis.Am J Public Health.1998;88:216-222.JMeliaTFrostRGraham-BrownProblems with registration of cutaneous malignant melanoma in England.Br J Cancer.1995;72:224-228.RASwerlickSChenThe melanoma epidemic: is increased surveillance the solution or the problem?Arch Dermatol.1996;132:881-884.RASwerlickSChenThe melanoma epidemic: more apparent than real?Mayo Clin Proc.1997;72:559-564.Accepted for publication October 22, 1999.Richard G. Roetzheim, MD, MSPH, was supported by a Generalist Physician Faculty Scholars Award from the Robert Wood Johnson Foundation, Princeton, NJ.Corresponding author and reprints: Daniel J. Van Durme, MD, University of South Florida, Department of Family Medicine, 12901 Bruce B. Downs Blvd, Box 13, Tampa, FL 33612 (e-mail: dvandurm@com1.med.usf.edu).

Journal

Archives of Family Medicine – American Medical Association

Published: Jul 1, 2000

References

  • Cancer Statistics Review: 1973-1989.
    Miller, B; Ries, L; Hankey, B; Kosary, C; Edwards, B
  • Malignant melanoma in the 1990s: the continued importance of early detection and the role of physician examination and self-examination of the skin.
    Friedman, RJ; Rigel, DS; Silverman, MK; Kopf, AW; Vossaert, KA
  • Cancer statistics, 1998.
    Landis, S; Murray, T; Bolden, S; Wingo, P